Upgrade Discs in NSA325-v2

schlaubstar
schlaubstar Posts: 4
edited January 2021 in Personal Cloud Storage
Dear community, I am struggling with a problem when upgrading my discs.

My initial situation: I have the NSA325-v2 with a RAID1 array on two disks with 2TB each. I will call this two disks O1 and O2 (Old 1 and Old 2). Everything is working perfectly but S.M.A.R.T has detected sector defects on O2 so I decided to upgrade.

To safeguard that disks are identical I bought two new Seagate SkyHawks with 4TB each (but listed on the compatibility list) which I will call N1 and N2 (New1 and New2).

What I try to achieve: Having a new RAID1 set with N1 and N2, containing all data of todays RAID set.

What I did:
  • replace O2 with N1, which lead to the NSA325 detecting a degraded RAID set which could be repaired
  • replace O1 with N2 .... which lead to the NSA325 to detect a RAID set that is down and cannot be repaired
I then went back to the combination of one O1 and N2 which lead to the detection of a degraded RAID set which could be repaired.
I then replaced the N2 with N1, which lead to the detection of a degraded RAID set which could be repaired.

I then hoped: Hey ... now I have the old data on both new disks so I should be able to insert them and have what I was looking for ... but all combinations with both new disks lead to the detection of a raidset which is down and can not be repaired. Each new disk individually will work as "mirror".
In all cases where both new discs are installed, they are recognized but can not be joined to a Raidset - the administration surface reccomends to add one disk to a JBOD collection, but never offers to add it to the raidset.

How can I accomplish, what I try to achieve?




Accepted Solution

  • Mijzelf
    Mijzelf Posts: 1,977  Guru Member
    Answer ✓
    Fishy. Have a look at the partition table on the new disks, is that GPT? If not, the box might choke on trying to offer an volume grow action.
    The box can work with both new disks inserted (apart from raid madness)? In that case it's not a power problem.

All Replies

  • Hi Mijzelf, thank you for your advise. To be honest: The new discs came from the factory so there was no Partition info on them.
    As I however tried a lot of constellations I used the Windows diskpart command to completely remove all partition infos from the disk. In that case, Windows > Computermanagement > Disk-Management just says: "A new disc has been added and needs to be initialized before you can access it". So I assume that there was no partition Info on the disc at all.

    I would have loved to use SSH but in case the raid-set is down, I do not see the menu-item "S.M.A.R.T" below the Volume and SSH seems not to start on the NAS, at least I can not connect.

    I will try to add GPT Partitionstyle to the discs and quick format them prior trying to plug them in and report about the progress. Thanks for the hint.
  • Mijzelf
    Mijzelf Posts: 1,977  Guru Member
    For SSH you'll need to install the SSH package. For SMART you need to install the SMART package.

    About the partitiontable, it doesn't matter what type of partition table the disk has when you put it in. The NAS will create a new one.
    My concern is that when you put a new 4TB disk next to an existing 2TB to create a raid1 array, it might be possible the NAS simply clones the 2TB disk, including its (MBR) partition table. In which case growing the array will not be possible.
    To see the partition table, log in over SSH (as root) and execute 'fdisk -l', When the table has a partition of type 'ee', there is a GPT table. Of course there is also a GPT table if fdisk tells you so.
  • Hi Mijzelf, thank you so much for your advise. It was the right answer.

    A few specific replies: I have the SMART and the SSH package installed, but when the NAS displayed the "raidset down" state, it did technically recognize the harddisk (as I was offered to add it to a JBOD array) but the SSH Server was not started (as I could not access the NAS via SSH) and the S.M.A.R.T Menu Item was missing (probably the service was also not started).

    I do not exactly know what was wrong in the first place as I accomplished to repair the raid array twice (from the good small disc to both of the large disks). I assume the partitions should have been correct at the latest, but the NAS was not capable to add both 'copies' to the raidset.

    However: It is fine now (except for the expansion which is currently pending). What I did was running windows: Deleting all partition infos via "dispart clean" - Then using the default "Computer Administration" - "Disk Management" which detected a new disk and asking which partition layout to add - replying with GPT. I did not create a partition or format the disk.
    After repeating the default procedure one more time (remove O2 and replace with N1, repair raidset, remove O2, replace with N2 ..) I was offered the option to repair the raidset again, instead of facing a "down" raidset like in the other attempts.

    As someone asked in another thread, how different states look like in the GUI, I will attach some screenshots illustrating the different stages (degraded raidset, recovery in progress, healthy again)

    Degraded raidset ("one disk is out of sync but NAS knows what the master is and can repair it")

    via SSH you see, that two disks are inserted but the second one is not part of the raid-set at that point of time

    After clicking "repair" the recovery process starts



    And finally the GUI and the mdadm view on the rebuild raidet